The dating of heated flint by TL has been used for several years in a wide range of situations. Many of the applications have allowed comparisons to be made between TL and other absolute dating methods. With very few exceptions, the consistency of the TL results with other dating evidence is found to be good. Other means of testing the technique suggest that TL dates on flint may be reliable over a time scale of over 400 ka.
This chapter begins with some general remarks about the application of TL to dating heated flint. These are followed by a section on the precision of TL flint dates and on the types of questions that can be answered by them. The next section sets out the criteria for the selection of samples for TL dating according to material type, visual appearance, size and burial situation. The procedures for carrying out the palaeodose measurement are next described. Finally, factors limiting the practical age range of the TL technique and methods for assessing the reliability of the dates are discussed.
2.1. General Considerations
The heating of flint in antiquity erases the TL signal which constitutes the memory of the material for its previous radiation exposure. By observing the present day TL intensity, a measurement can be made of the radiation dose absorbed by the flint since its heating. If the rate at which the radiation dose was delivered can also be measured, a means for determining the date of the initial burning is gained.
As discussed below, the age range over which TL dating of flint may be applied extends from the Iron Age to the Middle Pleistocene. The error limits on individual TL dates are typically around ±9%, but greater precision can be obtained by averaging measurements from a number of flints. By virtue of the material used, most applications are in archaeology rather than geology. They are particularly useful at sites that are beyond the range of radiocarbon dating, or that are lacking in datable organic matter which is associated with the occupation.
It has been said that TL dating of heated materials is one of the most direct means of dating site occupation. However, this statement needs some qualification. The event dated is the heating of the material to temperatures of 400°C or higher. This heating may well be the result of man-made fire, but it could also have a natural cause. Either way, it is certain that the flint would have to be on or very near ground surface for it to have reached such temperatures. Therefore it is more accurate to say that TL dates the exposure of the material on the surface, rather than its use. In a few cases, it has been clear that the TL dates do not refer to the manufacture and use of the flint artefacts, but to a previous or subsequent erosion of the material to the surface. Archaeological evidence of in situ heating, such as a localised distribution of burnt material, is therefore desirable, as it serves to associate the TL date with the archaeological context.
2.2. Precision of TL Dates on Flints
The error limits of TL dates are normally quoted as 1sigma limits, and should be interpreted as the limits within which there is a 68% probability that the true date lies. They quantify all sources of error, both random and systematic. It is this uncertainty that should be considered when comparisons are made with other dating methods or with TL dates produced by a different laboratory.
Error limits for flint TL dates usually lie in the range from ±7% to ±11%. The main factor affecting the size of the uncertainty is the environmental dose rate assessment. The gamma dose rate generally plays an important part in the age evaluation since it often dominates the total radiation dose received by the buried flint. Smaller error limits can result if the site has uniform radioactivity and has maintained an almost constant water content during its history.
When dating an assemblage of flints, it is usual to carry out TL measurements on several flints from the group. This allows random errors to be reduced, and leads to smaller error limits for the averaged measurement than can be achieved in the individual TL dates. Where the environmental dosimetry is not simple, multiple sampling can also lend confidence to the interpretation of the radiation measurements.
The above remarks consider TL as an absolute method of dating. However, in many applications the required information concerns the relative dating of two or more flint assemblages. Such questions are usually of the form: "Are these two flint scatters contemporaneous?" When answering this type of question, some of the systematic uncertainties can be taken out of consideration, and the date error limits are effectively reduced. The number of error sources that can be omitted depends on the relative positions of the two assemblages.
Where two flint assemblages derive from different sites, or are otherwise well separated, then the relevant error limits for dates measured by the same laboratory are reduced to typically ±7%. This would allow the dates of the clusters to be differentiated, with 95% success rate, if they are separated by a margin of 20%. If the two assemblages are more closely situated then further sources of error can be taken out of consideration. For instance, when two flint scatters are found on the same site and share similar burial conditions, the relevant error limits will reduce to typically ±5%. Under these conditions, the TL dates will have a 95% chance of discriminating between clusters that are separated in age by 15%.
2.3. Selection of Flint Samples for TL Dating
The following sections describe the main criteria for choosing samples which are most suitable for TL dating.
2.3.1 Suitable types of material
In these notes, the term "flint" is used as a broad category to mean any fine-grained siliceous material (including, for instance, quartzite). TL date measurements may also be performed on other types of heated stone, although the results may not be as precise or reliable as those on flint. In particular, difficulties are encountered when dating coarse-grained stones, because of the problem of zoning.
Zoning refers to the correlation of an inhomogeneous spatial variation of alpha radiation dose with a similarly inhomogeneous distribution of TL sensitivity. Alpha rays have a very short range (only 0.03 mm) in stone. Therefore, the presence of different minerals, having varying amounts of alpha activity, in the form of coarse grains can produce large spatial variations in the quantities of alpha dose deposited throughout the material.
Similarly, different minerals are likely to possess varying sensitivities to radiation, meaning that the efficiency in producing TL is also spatially variable on the scale of the grain size. When the TL measurement is performed, or the internal alpha activity of the stone is measured, only the spatially averaged values of the TL production or radiation dose rates are observed. As a consequence, the date calculations based on these observations will be in error, with the size of the error depending on the degree of correlation that exists between the distributions of radiation dose and TL sensitivity.
In the case of flint and similar materials, zoning is not a significant problem. The reasons for this are, firstly, that the fine-grained nature of these materials greatly limits the degree of inhomogeneity, and secondly, because the internal alpha dose typically makes a very minor contribution to the total radiation dose. However, in the case of coarse-grained stones neither of these conditions generally holds, and the possibility of a dating error is therefore significant. Nevertheless, the TL dating of stone can in many situations produce useful information.
2.3.2 Visual appearance of samples
Flints suitable for TL dating do not necessarily bear visible signs of having been burnt. The heating temperature required for initialising the TL (approximately 400°C) does not usually cause a colour change or surface cracking. Therefore, flints which are "unburnt" in the archaeological sense need not be rejected as potential TL dating candidates. The feasibility of a date measurement can be easily discovered by TL examination of a small (0.5 gram) portion of the sample. Such tests are routinely and rapidly performed in advance of a TL dating project.
If the temperature to which the flint has been heated is not quite sufficient to remove completely the pre-existing TL, it may be possible to recover information about the date of the heating by careful analysis of the measured TL data. However, in these cases, some loss of precision will be experienced, resulting in larger date error limits. Flints heated to very high temperatures are often best avoided as dating samples because TL sensitivity is reduced by excessive exposure to heat. This in turn may lower TL intensities to a level where they may be difficult to observe against the background of other luminescence processes.
2.3.3 Size range of flint TL samples
Before TL measurement, the outer surfaces of a flint sample are cut away in order to remove the part of the flint that has been exposed to the alpha and beta activity of the burial soil. This greatly simplifies the dose rate assessment. However, it also imposes a minimum size upon samples eligible for TL dating. Since a depth of 3 mm has to be cut away from all faces, the flint needs to be at least 8 mm across on its smallest dimension. Large flints should also be avoided, since their interiors may be significantly shielded from the gamma dose originating in the burial soil. Preferably, the smallest dimension of the flint will be less than 40 mm.
Exposure of the flint sample to light may potentially alter its TL measurement. However, the cutting away of the outer surfaces of the sample removes the parts which may have been thus affected. Consequently, it is not necessary to take extreme precautions to protect potential TL dating samples from light. Equally, it is advisable to avoid unnecessary exposures to very high light levels, particularly when the pieces are relatively translucent.
2.3.4 The burial environment - some considerations
As with all TL samples, the best flints for dating will derive from parts of the site where the burial conditions are uniform and have not altered. When considering the burial environment, it should be remembered that gamma rays have a range of 0.3 m in soil, and therefore all the deposits within this distance of the flint contribute to its radiation dose. It is easier to assess the radiation history of the flint if it was buried to at least 0.3 m depth in a relatively short time, and if its burial context was subsequently undisturbed. Note that long-term movements of water within the burial environment produce dose rate changes, and therefore constitute disturbances.
The radiation survey of the site, which routinely precedes most TL dating commissions, will reveal the most promising areas of the excavation where TL samples may be sought. The excavator will, in particular, be advised of any areas which are unfavourable for obtaining precise date measurements.
2.4. TL Measurement Procedures for Flint
TL is observed by heating the sample material in a controlled manner and recording the light emitted in a particular wavelength range. The graph describing the intensity of the light vs temperature is called a glow curve. After the flint's outer surfaces have been removed, the inner part is crushed. Grains of the crushed interior are laid on metal discs to be heated in the TL oven.
Figure 1. Glow curves of a sample of burnt flint, showing the characteristic peaking of TL intensity which occurs at the temperature of approximately 350°C. The glow curve of natural TL measures the present day TL intensity resulting from the flint's exposure to natural radiation during burial. Also shown are glow curves obtained after the flint has been given various laboratory radiation doses, measured in Grays (Gy), in addition to the natural dose.
Figure 1 shows a set of glow curves which were measured from several discs of a crushed flint. Some of the discs were irradiated with various known doses from a beta source, while the others were left unirradiated. The TL emitted by the unirradiated discs represents the signal due to the natural radiation dose which accumulated over the burial period of the flint, and is called the natural TL. All these measurements, of irradiated and unirradiated discs, are referred to as first glows. They clearly show the growth of the TL signal as increasing laboratory doses are added to the natural dose. The natural radiation dose, or palaeodose, is evaluated by extrapolating the growth of the TL backwards to zero intensity.
Palaeodose evaluation is complicated by the fact that the growth of TL vs dose is non-linear, and depends on knowing exactly how the TL grew in the dose region below the palaeodose. The form of the TL growth curve is determined by a further set of glow curve measurements. During the first glows, TL is erased from the flint so that a re-heating would produce no light emission. If, after their first glow, the discs are dosed with radiation, a TL signal will again be recorded when they are heated. These measurements are called second glows. The curve of TL growth vs dose in the second glow measurements provides the template for extrapolating the first glow data. Figure 2 compares the respective TL growth curves for first and second glows, and illustrates the extrapolation of the first glow measurements with a curve which has the same form as the second glow growth curve. The palaeodose is evaluated where the extrapolation curve meets the dose axis.
Figure 2. Curves describing the non-linear growth of TL intensity with increasing radiation dose. The lower curve is fitted to the second glow data points (crosses). The upper curve has the same form as the lower curve, but is fitted to the first glow data points (squares) which include measurements of the natural TL. The value of the palaeodose (or natural radiation dose) is obtained where this curve meets the dose axis.
2.5. Limitations on the Age Range of Datable Flints
The age range over which TL dating can be usefully applied is limited by various factors. The dating precision of very young flints is degraded by the presence of background luminescences which are not related to radiation exposure, and which are referred to as spurious signals. Very old flints may be difficult to date precisely due to saturation of the TL or its loss through decay while the flint was buried.
2.5.1 Spurious emissions from flint
Spurious luminescences form a background to the TL signal which can affect the palaeodose evaluation. For most flints of Neolithic or earlier periods these backgrounds are not of sufficient intensity to alter the measurements to a significant extent. However, for more recent flints, particularly those which have a low intrinsic TL sensitivity, or which have lain in a low radiation environment, the natural TL intensity can be very weak. In these cases spurious levels may not be negligible. However, by analysing the glow curve shapes it is often possible to remove the effects of the spurious emissions. When this is achieved, date measurements of useful precision may be obtained from flints which were heated as recently as the Iron Age.
In order to reduce spurious backgrounds to a minimum, sample discs are heated in an oxygen-free atmosphere. The unwanted luminescences are also reduced when the grain size of the sample is increased. These indications suggest that the origin of the spurious light is a chemical oxidation at the surface of the grains that occurs when heat is applied. To measure the effect of the spurious levels on palaeodose evaluation TL measurements are carried out on two very different grain sizes. If the results of these two sets of measurements are in agreement, then spurious luminescences are considered unimportant.
2.5.2 Saturation of the flint TL signal
At the other end of the age scale, TL dating precision is primarily limited by the effects of saturation. This term describes the existence of a finite limit to the quantity of TL which can be induced in a particular material. As radiation exposure increases, the latent TL signal does not grow indefinitely, but gradually approaches a limiting intensity. The TL growth curve is seen to bend over, until its slope vanishes and it becomes a horizontal line, indicating that further radiation exposure would produce no increase of the TL intensity.
In practice, the saturation TL intensity needs to be at least three times the natural TL level if the precision of the palaeodose evaluation is not to be adversely affected. The dose required to saturate the TL varies considerably between different samples of flint, but is typically in the order of thousands of grays. Clearly, flints exposed to low dose rates of natural radiation are least likely to encounter difficulties due to saturation. Because so much depends upon the individual flint and its burial situation, it is not possible to give a general rule about the maximum age that can be measured by TL. However, it would appear on present experience that ages of over 500 ka would be measurable in the majority of flints.
2.5.3 Instability of the flint TL signal
Another effect that needs to be considered when dating very old flints is that of TL stability. Over long periods TL signals decay, causing a reduction of the observed present day intensity. The lifetime of the TL signal is a characteristic of the signal, and does not vary between individual flints. If the age of the sample is short compared to the lifetime of the signal, no significant loss of TL occurs. If the sample's age is comparable to the signal's lifetime, a correction can be applied to the TL date to allow for signal decay (provided the signal's lifetime is known).
Ultimately, the only way of observing instability in a TL signal and determining its lifetime is to perform TL date measurements on "known age" samples. At present, the oldest flints for which comparisons exist between TL and other absolute dating methods are approximately 200 ka old. No evidence of signal decay is observed in these date measurements. Any limitation due to TL instability therefore remains a theoretical possibility rather than a demonstrated fact. The following section discusses the methods used in assessing the evidence for TL instability and other sources of error in TL dates.
2.6. Reliability of TL Dates on Flints
All date measurements depend on assumptions about initial conditions and subsequent developments which are often observationally inaccessible. The dating specialist is therefore reliant on a posteriori demonstrations for justifying the dating technique. The reliability of flint TL dates has been assessed in two ways; firstly, by comparison with other dating methods, and secondly, by inspection of results from sites in northern latitudes.
2.6.1 Comparisons with other dating methods
The rationale behind inter-technique comparisons is that, while individual dating methods may be in error in certain domains of age, sample type, or environmental conditions, etc., there is no expectation that these errors will be correlated. This assumes that there is no overlap between technical procedures, as exists, for example, between TL and ESR dating which share a common requirement for radiation dose rate assessment. In general, however, agreement between date measurements by two well-differentiated techniques will suggest that neither measurement is subject to error outside the quoted uncertainty.
At the most recent end of the age range of flint TL dates, several comparisons are available with radiocarbon dating of organic remains from related archaeological contexts. Generally good agreement is found between the two methods in the Holocene and for ages up to approximately 30 ka. However, at older sites the TL method tends to produce dates which are earlier than the corresponding radiocarbon ages. In this time range, the comparisons are complicated by an uncertainty in the correct calibration of the radiocarbon ages. It has been suggested that the problems of modern carbon contamination may be at least partly responsible for the discrepancies. Similar differences are also apparent between sediment TL and radiocarbon dates in the same age range.
For older sites, back to approximately 300 ka, useful comparisons can be made with the uranium series disequilibrium technique. The largest body of comparartive data exists at Oxygen Isotope Stage (OIS) 7. The data comprises TL dates on flint and both TL and uranium series results from stalagmitic calcite. Discrepancies between the methods do not exceed 15%, and may result from the normal inaccuracies in assessing the environmental dose rate in the TL method.
2.6.2 TL flint dates and Interglacials
As is well known, the Quaternary is characterised by a regular succession of climatic fluctuations which have been accompanied by large expansions and contractions of the glaciated regions. In northerly latitudes, human presence has been modulated by these changes, and has been mainly confined to the warmest climate phases.
For the dating specialist, this presents an ideal testing ground applicable to any absolute dating technique that covers a significant part of the Quaternary. If the dating method is reliable, the distribution of dates for sites in northern regions will reflect the approximately 100 ka periodicity with which the warm stages have recurred. If not, the distribution will appear uniform across the time range. In particular, the presence of a significant number of dates in the "forbidden" time zones would indicate technical problems.
The accumulated probability distribution of flint TL dates from sites in the British Isles has been calculated. In the Holocene, the distribution peaks between 7 ka BP and 8 ka BP, rising from zero at 13 ka BP, and falling again to zero at 2 ka BP. The fall-off reflects both the diminishing usefulness of TL dating for very young sites and the decline of the flint industry. The rising curve in the Early Holocene plausibly mirrors the increase of population during this time. Sparse occupation is known to have occurred in Britain in the Late Devensian; however, these sites are not represented among the TL dates.
No feature of the date distribution corresponds to OIS 5. Over the earlier warm phases, OIS 7, 9 and 11, the distribution curve represents only twelve TL dates from three sites. The dates from each site are mutually consistent, and, when averaged together, produce ages which are determined within uncertainties of between ±6% and ±10%. Encouragingly, these three mean dates correlate to a significant degree with the expected 100 ka periodicity. However, in view of the paucity of the data, the possibility of a fortuitous correlation cannot be confidently ruled out. As further data accumulate, it is certain that a clearer picture will emerge of the true reliability to be placed on flint TL dates.